[BUG]: runtime.node.event_loop.delay.avg metric disappeared

[EQuincerot]

Tracer Version(s)

5.41.1

Node.js Version(s)

v22.14.0

Bug Report

Since the v5.41.1, the metric runtime.node.event_loop.delay.avg disappeared from our dashboard.

The metrics comes back when we switch back to v5.41.0.

Reproduction Code

• monitor a node process with dd-trace-js
• add runtime.node.event_loop.delay.avg metric to a dashboard
• with dd-trace-js v5.41.0, the metric is visible in the dashboard
• upgrade dd-trace-js to 5.41.1, the metric is no longer sent
• downgrade dd-trace-js to 5.41.0, the metric is sent again.

Error Logs

No response

Tracer Config

No response

Operating System

No response

Bundling

Unsure

[Grmiade]

We have the same issue on our side.
We encounter some errors on the agent side when we try to send runtime metrics, like:

2025-03-11 13:49:51 UTC | CORE | ERROR | (comp/dogstatsd/server/server.go:623 in errLog) | Dogstatsd: error parsing metric message '"runtime.node.event_loop.delay.total:0[object Object]|c|#service:<service_name>,version:<version>"': could not parse dogstatsd metric values: strconv.ParseFloat: parsing "0[object Object]": invalid syntax

This also creates a memory leak on our services. Probably because the metrics are temporary stored in memory as long as they are not sent? By disabling DD_RUNTIME_METRICS_ENABLED, the memory leak is gone.

Could it be related to this recent change #5347?

[MMShep97]

I filed a ticket with their support staff; glad to see you might be experiencing the same thing @Grmiade (mem leak). I will link this thread there as well.

Don't think people can view this, but maybe for the devs: https://help.datadoghq.com/hc/requests/2068787

[EQuincerot]

We also had a memory recently, I suppose it's similar to #5360 but in my case, I don't have evidence that the memory leak comes from datadog.

[yo-shun-kou]

Tracer Version(s)
5.41.1

Node.js Version(s)
18.20.7

We also have a memory leak and got a lot of error messages about Datadog.

[Joosakur]

We also have the same error and the memory leak with 5.41.1. Had to revert back to 5.40.0.

[EQuincerot]

For us, it seems that memory leak arrived with 5.41.0 and the issue with metric with 5.41.1.
Do you confirm?

[danopia]

I'm seeing that 5.41.0 has some (possibly tolerable) memory impact, while 5.41.1 introduces excessive memory impact and also CPU impact requiring a rollback. Seems like there are two different issues at play.

[tommoor]

Memory usage after deploying 5.41.0 on our servers… honestly this version should be removed as it's a liability. Just glad I was able to find this thread